BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Machine translation from the user's perspective: what is it good f
 or? - Lucia Specia\, University of Sheffield
DTSTART:20121130T120000Z
DTEND:20121130T130000Z
UID:TALK40205@talks.cam.ac.uk
CONTACT:Ekaterina Kochmar
DESCRIPTION:After more than 60 years of research in machine translation (M
 T)\, the\nresulting technology has finally begun to be adopted for purpose
 s beyond\nmerely gisting. Many factors contributed for this significant ch
 ange in the\nway MT is perceived and used\, including noticeable improveme
 nts in\ntranslation quality. Evaluation campaigns (such as those organized
  by NIST)\nhave been showing steady progress over the years for certain\nl
 anguage-pairs. Industry-focused experiments have reported significant\ngai
 ns in productivity by using MT systems as part of their translation\nworkf
 low\, as well as significant cost reductions by using MT instead of\nhuman
 s for inbound translation (e.g.\, internal communication) or\nnon-crucial 
 content such as product reviews.\n\nThe decision on whether or not to adop
 t MT for a given purpose is directly\nconnected to the level of quality th
 at is expected by users. Different\ngroups of users prioritise different a
 spects of translation quality (e.g.\nmeaning over grammar). However\, a me
 tric of quality that accommodates these\ndifferent needs is yet to be defi
 ned.  In this talk I will introduce an\ninitiative towards formalising the
  notion of quality for human and machine\ntranslation as part of QTLaunchP
 ad\, an EC coordination and support action\,\nand describe some of my rece
 nt work on estimating some aspects of\ntranslation quality through automat
 ic evaluation metrics that do not rely\non human references.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR
